Our research study was conducted in Jaworzynka Valley in the Tatra Mts. in southern Poland – a mountain range located in central Europe and belonging to the Alpine mountain system (Fig. 1a, b). The process of reforestation and afforestation provides an opportunity to see how SOC accumulation has changed in calcareous soils in the Tatra Mts. due to different types of land use. Reforestation started in the Tatras in the 1960s when sheep grazing was banned in the interest of nature conservation. Since that time, some abandoned pastureland has undergone natural succession, while some has been afforested as part of a major government afforestation program in Poland.

Animals (mainly sheep) had grazed in Jaworzynka Valley since the 16th century (Fig. 2). Grazing was banned in 1963. The soils on the sides of the valley became heavily eroded. In light of the risk of further erosion, dwarf pine shrub (Pinetum mughi) and larch (Larix sp.) forest were planted (1962–1963). Afforestation efforts stopped in 1970. Some places have been left as mountain grasslands (clearings). Jaworzynka Valley now features a mosaic of the aforementioned plant communities (Figs. 1b, 3). No forest management work has been done since the ban on grazing and the start of afforestation, and all plots undergo natural succession as part of a national park.

The valley is made up of dolomitic limestone (Anisian, Ladinian) (Sokołowski and Jaczynowska, 1979). The mean annual air temperature in the study area (data for the nearest station: Hala Gasienicowa) is 2.4∘ C. The mean annual temperature of the warmest month (August) is 10.8∘ C (min. -4.1∘ C; max. 17.6∘ C). The mean annual temperature of the coldest month (February) is -5.3∘ C (min. -21.4∘ C; max. 7.2∘ C). The valley's mean annual precipitation is 1661 mm (Błażejczyk et al., 2013).

The research study was conducted on three plots (30 m × 30 m) located close (Fig. 1b) to each other to avoid differences connected with slope position and exposure, which can affect SOC stock variability (Fernández-Romero et al., 2014) as well as geological and soil heterogeneity that can affect SOC stocks and forms of OM (Baldock and Skjemstad, 2000; Parras-Alcántara et al., 2014). Plot no. 1 represented a mountain grassland (high mountain calcareous grassland: Carici sempervirentis–Festucetum tatrae association). Plot no. 2 represented thickets of dwarf pine Pinetum mughi. Plot no. 3 represented sparse larch (Larix sp.) forest (ca. 400 trees ha-1) with a dense cover of grass Calamagrostis sp. on the forest floor (Fig. 4). The three plots were located at an elevation between 1200 and 1220 m (location: 49∘15′32′′ N, 19∘59′35′′ E) on a uniformly inclined slope (linear mountain side slopes) of 25∘ SW.

Nine soil profiles were excavated in each plot; therefore, the soil material was collected from 27 pits. Soil profiles were excavated down to the lithic contact and described according to Schoeneberger et al. (2002). A reference profile was selected at each study site – the soil profile nearest to the central point of the plot (Table 1).

Unlike in the case of many other research studies, it was decided to collect samples from the genetic horizons of the soil, and not from the intervals. Organic O horizons are well-developed in mountain soils (Kubiena, 1953; Drewnik, 2006); therefore, it is necessary to take into account the boundary between the organic O horizon and the A horizon in research focused on the mechanisms of OM storage in soil.

One large (mineral sample – approx. 2 kg, organic sample – approx. 0.3 kg; moist), representative bulk sample was collected from each studied genetic soil horizon, then placed in sterile polyethylene bags. In addition, undisturbed soil samples were collected from the reference soil profiles in order to determine the bulk density of fine soils. In this case, due to a very large quantity of stone and gravel, a steel frame (20×20 cm) was used to obtain a large sample in the form of a rectangular prism, with a volume ranging from 4000 to 8000 cm3, depending on the horizon. The studied soils were classified according to the WRB system (IUSS Working Group WRB, 2007).

Bulk soil samples taken from A horizons and mineral B horizons were air-dried, gently crushed using a wooden rolling pin, and sieved using a 2 mm sieve. Live roots were removed. Soil samples from O horizons were milled after the living parts of plants in the samples had been removed.

Stone and gravel content (particles > 2 mm) was determined by weighing. Bulk density of fine soil (BD(f)) (mass volume-1) was calculated according to Don et al. (2007) as follows: BD(f)=masssample-massparticles>2mmvolumesample-massparticles>2mmdensityparticles>2mm. Texture was determined by wet sieving (sand fractions) and the hydrometer method (silt and clay fractions) (Gee and Bauder, 1986). The concentration of total carbon (TC) and nitrogen (N) was determined by dry combustion gas chromatography using a CHNS analyzer (Elementar vario MICRO cube elemental analyzer). CO2 content obtained from carbonates (CO2(carb)) was determined using the volumetric calcimeter method. Each sample's pH was measured in deionized water (1:2.5 soil / water ratio) (Thomas, 1996). Soil color was described in the moist state using Munsell Soil Color Charts (Oyama and Takehara, 2002).

SOC was calculated by subtracting inorganic carbon (SIC – carbon from CO2(carb)) from TC. SOC stocks were calculated according to Don et al. (2007), as follows: SOC=∑t=0nBD(f)×SOC×depthvolume. In the equation, “depthvolume” is the depth of the horizon minus the volume of particles Ø > 2 mm. This calculation excludes particles Ø > 2 mm as they are not a component of bulk density.

Physical fractionation of the soil was carried out according to the Leifeld and Kögel-Knabner (2005) method to obtain several OM fractions: free light particulate fraction (POM LF1), light fraction occluded in macroaggregates (POM LF2), residual fraction occluded in microaggregates (ROM), and MOM: MOM fraction outside water-stable aggregates (MOM FF1), MOM fraction inside macroaggregates released after their disruption (MOM FF2), and MOM fraction inside microaggregates released after their disruption (MOM FF3). A 30 g sample of air-dried soil (< 2 mm) was immersed in deionized water on a 20 µm mesh. After 5 min, the material was gently sieved to obtain a mineral fraction outside water-stable aggregates < 20 µm (FF1). The sieved material was air-dried at 40∘ C and weighed. It was then transferred into a 100 mL centrifuge beaker and sodium polytungstate solution (1.8 g cm-3) was added. The resulting material was gently stirred, left for 10 min to settle, and centrifuged for 10 min (2320 g).

POM LF1 was decanted and washed with deionized water using a 20 µm mesh. The residual soil material was dispersed ultrasonically (Sonics 750) with an energy of 22 J mL-1 to break down macroaggregates. The dispersed soil was sieved at 20 µm mesh to obtain a mineral fraction < 20 µm inside macroaggregates (FF2), dried, and POM LF2 was separated as described for POM LF1. The residual soil material was dispersed ultrasonically with an energy of 450 J mL-1 to break down microaggregates. The material was sieved using a 20 µm mesh to obtain a residual fraction (ROM) and obtain a mineral fraction < 20 µm inside macroaggregates (FF3). The mineral fractions < 20 µm (FF1, FF2, FF3) were collected after each dispersion step, washed by centrifugation, and weighed. The mass of these fractions relative to the total mass of the < 20 µm fraction was taken as a sign of soil aggregate stability.

The concentration of TC, N and CO2(carb) as well as the content of particles Ø > 2 mm were determined for 106 samples from all the studied soil profiles, while the texture, bulk density, pH, and OM fractionation were determined for samples from reference profiles (Table 1). Only material from A horizons was fractioned, because the soil material from O horizons was poorly decomposed without signs of mixing with mineral matter.

In this study, it was considered justified to use descriptive statistics for presenting the value structure of variables resulting from measurements. The use of statistical inference to assess the significance of differences between the characteristics of three different sampled areas was not considered warranted, given the inadequate number of samples (3×9 profiles).

According to the WRB system (IUSS Working Group WRB, 2007), reference profiles no. 1 and 3 were classified as Rendzic Hyperskeletic Leptosols (Humic, Eutric), while profile no. 2 was classified as a Folic Hyperskeletic Leptosol (Calcaric, Humic) (Table 1).

O horizons (Oi) in profiles no. 2 and 3 had a thickness of 20 and 2 cm, respectively. The content of Ø > 2 mm particles increased with depth from 0 to 70–90 % at a depth of 30 cm. Fine soil had a dark color in the range of 10YR 2–4/1–4 and silt loam texture (Table 1). The studied soils were characterized by a very low bulk density of approx. 0.02–0.03 Mg m-3 in O horizons and approx. 0.18–0.43 Mg m-3 in A horizons and B horizons. Carbonates were found to be present in the fine soils. Carbonate content was very low (0.00–33.5 g kg-1 of carbonate CO2) in organic O horizons and increased with depth from 250 g kg-1 of carbonate CO2 in A horizons to 300–400 g kg-1 of carbonate CO2 in B horizons (Table 1). Soil pH values change with depth. The pH value range measured in distilled water was 4.3–5.1 (profile no. 2) and 6.3 (profile no. 3) in O horizons, 7.5–7.7 in A horizons, and approx. 7.8 in B horizons (Table 1).

Mean SOC concentration was the highest in O horizons (465.3 g kg-1 in soils under dwarf pine and 351.7 g kg-1 in soils under larch forest), medium in A horizons (56.8–65.5 g kg-1), and the lowest in B horizons (15.9 g kg-1 under larch forest and 21.7 g kg-1 under grassland) (Table 2).

The SOC stock was 63.5 Mg ha-1 in soil under larch forest, 47.5 Mg ha-1 in grassland soil, and 42.9 Mg ha-1 in soil under dwarf pine shrub (Fig. 5). In all plots, SOC was accumulated mainly in the mineral part of soil (A horizons and B horizons): 100 % of SOC in grassland, 94 % in larch forest, and 67 % in soil under dwarf pine.

The C / N ratio varies depending on the type of soil horizon. It was 41.9 in the O horizon in soils under dwarf pine shrub, and 28.5 in soils under larch forest, and ranged from 10.7 to 13.5 in A horizons and B horizons in all the studied soils (Table 2).

The mass of the FF2 fraction accounted for more than 60 % of the < 20 µm fraction total in soils under dwarf pine shrub and larch forest and more than 52 % of the < 20 µm fraction total in soil under grassland (Table 3). The mass of the FF3 fraction of the < 20 µm fraction total ranged from 27.9 to 31.6 % in soil under dwarf pine shrub and larch forest, and from 43.6 to 44.8 % in soil under grassland. The mass of the FF1 fraction of the < 20 µm fraction total was the lowest in all the studied soil profiles (accounted for 0.9 % in the A2 horizon in grassland soil to 9.6 % in the A1 horizon in soil under dwarf pine).

In A horizons, SOC content in MOM fractions accounted for 67.6 to 85.8 % of total SOC, while in POM + ROM fractions, it ranged from 12.4 to 32.4 % of total SOC (Table 4).

The SOC concentration of the two POM fractions (POM LF1 and POM LF2) ranged from 227.4 to 308.3 g SOC kg-1 (Table 4). The highest content of SOC of POM LF1 was found in the A1 horizon of soil under dwarf pine shrub (10.3 % of SOC), and the lowest in the A2 horizon in soil under grassland (1.7 % of SOC). SOC content of POM LF2 was the highest in the A1 horizon in soil under dwarf pine shrub (18.3 % of SOC) and the lowest in the A2 horizon in soil under larch forest (5.3 % of SOC) (Table 4). SOC concentration in MOM fractions ranged from 35.6 to 97.1 g SOC kg-1. The SOC content in MOM FF2 was the highest among all MOM fractions, and ranged from 37.9 % of SOC in the A1 horizon in grassland soil to 59.2 % of SOC in the A2 horizon soil under dwarf pine shrub. SOC content in MOM FF3 accounted for 15.3 % of SOC in the A2 horizon under larch forest to 27.1 % of SOC in the A1 horizon in soil under grassland (Table 4). SOC content in MOM FF1 was the lowest among all MOM fractions and accounted for 0.5 % of SOC in soil under grassland to 3.8 % of SOC in soil under dwarf pine shrub.

The C / N ratio varied depending on the OM fraction from 17.1 to 34.0 in POM fractions and from 7.2 to 11.5 in MOM fractions (Table 5).

The studied soils possessed characteristics of soils formed on carbonate rock in a fairly harsh mountain climate (Jenny, 1930; Kubiena, 1953; Skiba, 1983). The soils had a high SOC concentration in A horizons, and thick O horizons occurred in soils formed under dwarf pine shrub (Bochter and Zech, 1985; Drewnik, 2006). The studied soils had a high pH due to their calcareous parent material despite the humid climate in the region (Table 1). However, despite their typicality, these soils were much shallower and had a larger content of stone than soils formed under comparable conditions in the Tatra Mts. (Skiba, 1983; Miechówka, 2000; Drewnik, 2006), which is most likely due to the fact that these soils occur on slopes, which are subject to strong erosion caused by animal grazing. Very low bulk density occurred in O horizons in comparison with other results from mountainous soils (Kammer et al., 2009; Budge et al., 2011; Martinsen et al., 2011) and this can be explained by the fact, that these horizons are built of relatively fresh, loose material, with no signs of advanced humification.

The effect of vegetation type in the Jaworzynka Valley can be observed in terms of changes in OM content in the soil. The greatest accumulation of SOC was observed in soils under larch forest, and less accumulation was observed in soils under grassland and under dwarf pine shrub (Fig. 5). The highest stock of SOC sequestered in the A horizon in soil under larch forest can be explained by the nature of local vegetation. The larch forest was sparse with a dense cover of Calamagrostis sp. on its floor (Fig. 4). No biomass measurements were done as part of the study, but it may be assumed that tall Calamagrostis sp. grass supplies a higher amount of plant tissue to the soil than short xerothermic grassland plants do. Guzman et al. (2014) found that after 3 years of the reclamation of mining area soils using tall prairie grasses, these soils had received significantly more biomass than soils reclaimed using cool-season forage grass. In the Jaworzynka Valley, the relatively higher SOC stock in larch forest (Fig. 5) can be also explained by a specific forest microclimate affecting the composition of forest floor species and OM decomposition rates and the fall of needles, which does not occur in grassland areas (cf. Kim, 2000; Seeber and Seeber, 2005; Rigueiro-Rodríguez et al., 2012). Soils covered with larch forest are not studied very often; however, the SOC stock in Jaworzynka Valley soils in plot no. 3 was in the range given for soils found in larch forests (formerly agriculturally used Luvisols) in China (Wang et al., 2011).

While the SOC stock was quite uniform throughout the entire soil profile in soils found under dwarf pine shrub as well as soils under grassland, it was grassland soils that had a significantly higher amount of SOC sequestered in the studied A horizon rather than dwarf pine shrub soils (Fig. 5). This finding is partly consistent with other newer results cited by Laganière et al. (2010) and Poeplau and Don (2013), providing evidence of decreasing SOC stocks following grassland reforestation in the first several years after the conversion from pastureland to grassland. This is most likely due to certain constraints placed on the delivery of very rapidly decomposable grass roots in forest ecosystems in comparison with grassland ecosystems (Oades, 1988).

In coniferous forests ecosystems, thick O horizons develop (Table 1) as a result of a large amount of litter supplied to the soil as well as a specific forest microclimate and acidification caused as a result of needle decomposition (Bochter and Zech, 1985; Seeber and Seeber, 2005; Drewnik, 2006). The development of O horizons can rarely offset SOC depletion caused by the limited supply of grass roots (Kammer et al., 2009). In the studied soils under dwarf pine shrub, the SOC stock in the O horizon comprised only one-third of the entire SOC stock despite its substantial thickness, which is similar to the results obtained in forest soils in the Stołowe Mts. in southwestern Poland (Gałka et al., 2014).

The relatively small SOC stock in dwarf pine shrub soils may be the result of the fact that it is a relatively young plant community and large amounts of OM – derived from the roots of dead trees, as observed by Debasish-Saha et al. (2014) in the soils of the Lower Himalayan hills – have not yet appeared.

The SOC stock determined for individual plots in our study area (Fig. 5) was significantly lower than that found in similar mountain environments in the temperate climate zone (Kammer et al., 2009, Gałka et al., 2014). The most important reason for the relatively small SOC stock was the small depth of the soil and a very large number of Ø > 2 mm particles (Table 1). This produced a low SOC stock despite a high content of these elements in the local soil mass (fine particles) (Table 2). Studies on SOC stocks have not been carried out in the Tatra Mts. so far, but results obtained by Skiba (1983) and Miechówka (2000) indicate that soils developed on calcareous parent material and under similar conditions are characterized by a similar or higher concentration of SOC, greater thickness as well as a smaller number of Ø > 2 mm particles, which allows to suppose that the SOC stock would also be higher in typical Tatra soils than that in the studied soils.

The lower SOC stock determined for the studied soils in comparison with Haplic Cambisols in the Urals (Kammer et al., 2009) can be explained by greater biological activity in calcareous soils or more favorable climate conditions devoid of drought periods. The SOC stock determined for our study area was significantly smaller than that found in a similar environment in the Stołowe Mts., which are part of the Sudeten Mts. in southwestern Poland (Gałka et al., 2014). This can be explained by the fact that the soils in the studied area were much more shallow.

In the studied soils, OM was mainly associated with the mineral part of the soil (Table 4), which is consistent with research results obtained for other soils in temperate climate zones, where more than 60 % of the total SOC fraction is associated with the mineral part of soil (Jastrow, 1996; Don et al., 2009). OM in the soils studied can be divided into two main groups: (1) MOM FF fractions (MOM FF1, MOM FF2, MOM FF3) with a low C / N ratio suggesting a relatively high contribution of humified as well as microorganisms-derived OM, which is believed to serve as a factor that glues soil particles together inside microaggregates (Oades, 1984; Six et al., 2001), and (2) POM fractions (POM LF1, POM LF2) with a high C / N ratio, which suggests relatively weakly decomposed plant-derived material (Table 5).

The highest mass of FF2 fractions in relation to the mass of the FF1 fraction in all the studied soils (Table 3) refers (Leifeld and Kögel-Knabner, 2005) to the relatively high structural stability linked with the development of water-stable aggregates in the studied soils, in comparison with sandy soils investigated by Leifeld and Kögel-Knabner (2005). This can be an effect of the calcium and magnesium carbonate content as well as the relatively high content of mineral colloids, which promote structural stability in soils (Oades, 1988; Muneer and Oades, 1989; Denef et al., 2004; Lützow et al., 2006; Grünberg et al., 2013). Similarly, in the studied soils, OM associated with the mineral part of the soil occurred mainly in macroaggregates (MOM FF2) and microaggregates (MOM FF3) (Tables 3, 4). In all the studied soils, the amount of SOC outside of water-stable aggregates (MOM FF1) was very small. The highest concentration of SOC in macroaggregates in the studied soils was consistent with results obtained by Jastrow (1996) and Debasish-Saha et al. (2014). The researchers independently concluded that this is the result of the transitional nature of macroaggregates, in comparison with microaggregates, which contain a relatively passive pool of OM because of strong bonds with clay minerals.

In soil under grassland, the largest amount of SOC bound within MOM FF3 (microaggregates OM) was found in contrast to soils found under coniferous communities (dwarf pine shrub and larch forest) (Table 4). This corresponds with the highest mass of FF3 and the lowest mass of FF1 and refers (Leifeld and Kögel-Knabner, 2005) to high structural stability linked with the development of stable microaggregates in soils under grassland. Conversely, the mass of the FF1 fraction was the largest in soils found under dwarf pine, while the largest part of SOM was stored in the MOM FF2 fraction. This shows that the soil structure was less stable (quantity of FF1 fraction) and most of the SOM occurred in more transitional aggregates (quantity of FF2 fraction) (Leifeld and Kögel-Knabner, 2005; Debasish-Saha et al., 2014).

The C / N ratio in MOM did not change with depth along the soil profile (Table 4). The lowest C / N ratio was observed in the MOM FF1 fraction, which was most likely due to less protection of OM from microbial attack outside water-stable aggregates (Lützow et al., 2006).

A much higher content of the POM LF1 and POM LF2 fractions in the soil under dwarf pine shrub versus that in soil under grassland and under larch forest was observed, both outside aggregates (POM LF1), as well as occluded in macroaggregates (POM LF2) (Table 4). The highest amount of the POM LF1 fraction seemed to confirm suppressing decomposition rates in soils under coniferous vegetation in contrast with grassland soils. This trend can be affected by the suppressing effect of coniferous plant material (lowered pH) on soil microbial activity (Drewnik, 2006) as well as lower soil temperature that limits soil biological activity in forest ecosystems (Kim, 2000) or by the delivery of more recalcitrant plant material such as tree roots to the forest soil (Debasish-Saha et al., 2014). Budge et al. (2011) found – in Alpine soils – that the residence time of POM derived from dwarf shrubs was longer than the residence time of POM derived from grassland, although researchers have not settled upon whether this is the result of the suppressing effect of the plant community on microbial activity or if it is the result of higher recalcitrance of the material delivered to the soil.

The obtained results appear to confirm a very important role of the POM fraction as a precursor of macroaggregates (Lützow et al., 2006; Debasish-Saha et al., 2014; van Leeuven et al., 2015). OM was supplied to the soil functions as a nucleon-gluing soil colloid due to soil microbial activity. Results obtained by Debasish-Saha et al. (2014) show that the higher amount of the POM fraction in forest soils – in comparison with soils developed via other types of land use – directly leads to a higher abundance of macroaggregates in these soils. Van Leeuven et al. (2015) as well as Tejda and Benítez (2014) observed that OM delivery to soils found under meadows as well as to eroded agricultural soils causes a higher biomass of soil bacteria and fungi. In the special case of Icelandic meadows studied by van Leeuven et al. (2015), it leads to a higher mass of macroaggregates in these soils.

OM occurring outside aggregates (POM LF1) was less decomposed than OM occluded in macroaggregates (POM LF2) in all the studied soils, as evidenced by the highest C / N ratio (Table 5). This proves that relatively young organic material (poorly humified) is gradually incorporated into soil aggregates (cf. Oades, 1984; Six et al., 2001; Lützow et al., 2006; Budge et al., 2011). The C / N ratio was observed to increase in the POM LF1 fraction with depth in all the studied soils (Table 5). This suggests that the primary precursors of POM formation are plant roots because the decomposition rate of OM increases with depth within O horizons developed as a result of litter accumulation on the mineral surface of the soil (i.e. C / N ratio decreases) (Ussiri and Johnson, 2003; Budge et al., 2011).

In contrast, the C / N ratio in the POM LF2 fraction decreased with depth (Table 5), which indicates a higher decomposition rate for OM occluded in aggregates with depth. This can be the result of incorporation of weakly-decomposed, relatively fresh OM into the structures of aggregates, which occurred most rapidly in the top of the mineral part of soil (A1 horizon) in comparison with deeper horizons. In addition, this suggests the action of mechanisms that incorporate POM into soil aggregates found near the soil surface. This can include significant activity of soil fauna, which is abundant in calcareous soils (Zanella et al., 2011) in the top soil horizons and increased activity of microorganisms that live near the soil surface (Beier and Rasmussen, 1994).

Research in the Jaworzynka Valley showed which method of renaturation is the most favorable in shallow eroded calcareous soils developed on steep slopes found in a temperate humid climate. This yields the following question: which solution is more advantageous? (1) Allow grassland to remain grassland; (2) carry our afforestation using sparse larch forest with a naturally developed dense cover of tall grass Calamagrostis sp. on the forest floor; and (3) plant dwarf pine shrub. All three methods effectively protect the soil against erosion, which was confirmed by the horizonation of all the studied soils – seen to be experiencing gradual soil stabilization (Targulian and Krasilnikov, 2007; Guzman et al., 2014).

The expected effect of soil reclamation and renaturation is SOC (OM) sequestration, especially in resistant (mineral-associated, aggregated) forms, which leads to increased soil fertility, soil stability, and improved soil structure (resistance against erosion), and helps restrict CO2 emission into the atmosphere (Oades, 1984; Debasish-Saha et al., 2014; Fernández-Romero et al., 2014; Guzman et al., 2014). In this context, afforestation with sparse larch forest is the most effective solution, because soils under larch forest sequester the largest amount of SOC (Fig. 5), and a significant part of OM occurs in aggregates resistant to mechanical degradation and decomposition, especially in contrast with dwarf pine (cf. Oades, 1984; Leifeld and Kögel-Knabner, 2005; Debasish-Saha et al., 2014). In soils found under meadow, most SOC is sequestered in mineral-associated form. This is advantageous in terms of SOC sequestration itself, but can be less favorable in terms of soil erosion, as smaller aggregates can be relatively easily moved by the wind and water as well as from the point of view of water and air retention and microorganism habitats in the soil (Oades, 1984; Barthès and Roose, 2002). In this context, the planting of dwarf pine and larch forest would be more beneficial.

Taking all relevant information into account, it appears that the planting of larch forest would be the optimal solution in the context of SOC sequestration, improvement of soil properties, and protection against erosion. Experimental data suggest that mixed-type vegetation such as grass on the forest floor protects the soil against erosion more effectively than forest vegetation with a bare forest floor and meadow does by itself (Mekonnen et al., 2014).